#
✕ nick:rqou_
#asahi-dev · 2022-03-28 02:26
rqou_isn't that supposed to be the A15?
#asahi-dev · 2022-03-28 02:26
rqou_decided to start taking a peek at the prores encoder, and does anybody know what changed in a dart,t8110? or why the M1 Max would have that in the ADT?
#asahi-dev · 2022-03-27 18:35
rqou_alyssa: yeah, "R" is definitely the "real name" people IRL know me by, even if it may or may not be my wallet name. and as for "a court should be able to find you", given the email address i'm currently submitting under, i'm clearly not trying to be anon, names are just hard
#asahi-dev · 2022-03-27 11:08
rqou_nah names are just hard
#asahi-dev · 2022-03-27 11:02
rqou_i literally have everybody even irl call me that, so ¯\_(ツ)_/¯ . you can resubmit it under your name i guess?
#asahi-dev · 2022-03-27 11:00
rqou_but in any case, i don't really consider anything in those three commits to even be copyrightable
#asahi-dev · 2022-03-27 11:00
rqou_i can add a signoff though
#asahi-dev · 2022-03-27 10:59
rqou_uhh, the real name i go by nowadays is just R
#asahi-dev · 2022-03-27 10:51
rqou_so that i can actually submit proper PRs
#asahi-dev · 2022-03-27 10:51
rqou_working on fixing my setup rn
#asahi-dev · 2022-03-27 10:51
rqou_sorry for how wonky it is, i don't have a full clone of linux with all the history
#asahi-dev · 2022-03-27 10:50
rqou_anyways, code here: https://github.com/rqou/linux/commits/atc-m1max
#asahi-dev · 2022-03-27 10:50
rqou_sven: yes. i eventually realized that i can just set the size to 8 instead of 4 to make it work
#asahi-dev · 2022-03-27 10:07
rqou_now i just need to figure out how to fix the efuses that don't align with the word size, otherwise the port on the right hand side doesn't work
#asahi-dev · 2022-03-27 10:05
rqou_oh it was a bug in sven's code. anyways, working USB3 on M1 Max now!
#asahi-dev · 2022-03-27 09:37
rqou_i must be missing something simple
#asahi-dev · 2022-03-27 09:36
rqou_the fuses fail to load
#asahi-dev · 2022-03-27 09:36
rqou_ok, i've found the right addresses to plug in to instantiate the atcphy, and i've filled in the efuses, and yet still nothing happens
#asahi-dev · 2022-03-27 08:41
rqou_poking sven again?
#asahi-dev · 2022-03-27 08:41
rqou_hmm i also just realized that i don't know which address blocks in the ADT correspond to the address blocks i would need to put into the dtsi for ATCPHY
#asahi-dev · 2022-03-27 02:22
rqou_that's the name in the debug message, and the atc-WIP branch uses very similar names
#asahi-dev · 2022-03-27 02:15
rqou_in any case, here are the efuse addresses for the tunables: https://gist.github.com/rqou/153d958648987053aed474731b9f8156
#asahi-dev · 2022-03-27 02:11
rqou_cc sven?
#asahi-dev · 2022-03-27 02:10
rqou_hi, i reverse engineered the ATCPHY efuse settings for the M1 Max and am trying to figure out how to plug it into the code in the atc-WIP branch, but i'm not sure how to encode in the DT the settings that straddle two different efuse words
#asahi-dev · 2022-03-25 08:07
rqou_but has been hacked up with tweaks as i experimented with what various registers do, so it's now a complete mess
#asahi-dev · 2022-03-25 08:06
rqou_the script mostly comes from driver RE
#asahi-dev · 2022-03-25 08:06
rqou_because they all interact with each other in complicated ways
#asahi-dev · 2022-03-25 08:06
rqou_the parts that are most dependent on driver RE have to do with calculating various parameters related to macroblock sizes/strides and chroma subsampling
#asahi-dev · 2022-03-25 08:05
rqou_it is a mix of driver RE, hypervisor traces, and ad-hoc manual poking of the hardware
#asahi-dev · 2022-03-25 07:57
rqou_i have read and understood the RE policy, and my current WIP notes are stored in a github gist at https://gist.github.com/rqou/2dafd40cfe0362cc84c3ee26c68b2b36
#asahi-dev · 2022-03-25 07:56
rqou_hi, i've been doing a bunch of reverse engineering of the JPEG encode/decode hardware, and i currently have it figured out to the point where i have proofs of concept of encoding/decoding, and i've started writing up documentation of my current understanding of the hardware. what can/should i do next in order to enable other people to start poking at it and possibly write drivers for it?